用统计方法自动决定哪些语言该翻译，提升多语言文本分类准确率。
Discovering Translation-Worthy Languages with E-Values
- 基于配对e过程动态比较直接分类与翻译后分类效果。
- 在SIB-200和MASSIVE上分别提升准确率8.14和16.70个百分点。
- 结果稳定可审计，适合需要高可靠性多语言系统的研究者。
选择何时翻译多语言文档是文本分类中的核心路由问题：翻译可能提升某些语言的预测效果，却会降低其他语言的性能或增加不必要的计算开销。统一翻译或启发式语言层级无法提供统计控制的路由决策。本文提出一种基于配对e过程的语言级路由机制，能持续比较直接分类与翻译辅助分类的效果，并在确定后冻结路由策略。通过家族误差控制的阈值280，将每数据集中14种可选语言的任意错误路由概率控制在0.05以内。在SIB-200和MASSIVE数据集上，该路由器分别为15种语言中的4种和15个语境中的14个选择翻译，相比直接分类分别提升了8.14和16.70个百分点的置信度外准确率。所有28项决策在50次独立排序及组内阈值下均保持稳定。结果表明，配对e过程可实现统计可控、随时有效且可审计的多语言分类路由。
原文摘要 · Abstract (English)
Choosing when to translate multilingual documents is a central routing problem in text classification: translation can improve predictions for some languages while degrading others or adding unnecessary computation. Uniform translation and heuristic language tiers do not provide statistically controlled route selection. We introduce a language-level router based on paired e-processes that continuously compares direct and translation-assisted classification before freezing a routing policy. A familywise-controlled threshold of 280 bounds the probability of any false route across 14 eligible languages per dataset by 0.05. On SIB-200 and MASSIVE, the router selects translation for 4 of 15 languages and 14 of 15 locales, improving held-out accuracy over direct classification by 8.14 and 16.70 percentage points, respectively. All 28 decisions remain stable across 50 outcome-independent orderings and relative to the per-group threshold. Our results demonstrate that paired e-processes enable statistically controlled, anytime-valid, and auditable multilingual classification routing.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。